Evaluation rankings评测排行
Scores you can play.分数之外，游戏也能亲手验证。
Every row links to the games, failures, clean source, and exact evaluation record. Results from different benchmark releases are never mixed. 每条成绩都能打开对应游戏、失败项、整理后的源码和完整评测记录。不同测试集版本的结果不会混在一起排名。
v0.4.0 · current
GameBench improves through versioned releases. This page keeps scores comparable by showing one ruleset at a time. GameBench 会持续迭代。每个页面只比较同一版测试规则下的成绩，避免不同版本混排。
Experimental
Current measured results当前阶段性成绩
0 active results on this exact ruleset.条结果采用这一版测试规则。
No stage result is available for v0.4.0 yet. No active stage result has been published for this version.
Official
Official qualification正式成绩资格
No Official result yet. A model must complete every required game under three fixed random conditions, then pass clean rebuild and independent verification.目前还没有正式成绩。参评模型需要在三组固定随机条件下完成全部游戏，并通过无密钥重建与独立复核。